MERCURY in FISH What Fish Are Safe?

Home , Mercury in fish, Tilapia

UCSC SIGMA XI LECTURE 2009 Tuesday April 28 -- 8 pm Engineering Lecture Hall, UCSC MERCURY IN FISH What Fish Are Safe? NED GROTH Senior Scientist, Consumers Union (retired), consultant to UN Food and Agriculture Organization and World Health Organization

The speaker, Dr. Ned Groth III, has recently authored a major study on methyl mercury in fish. This talk will dis- cuss cases of people who suffered from methylmercury poisoning after eating widely consumed fish. Methylmer- cury is very toxic to the nervous system. Some kinds of fish have much more methylmercury than others, and some people are much more sensitive to mercury than others. The talk concludes with advice on which fish are safe, which fish should be consumed in small quantities, and which fish should be eaten rarely if ever. It should be of interest to everyone who eats fish or is interested in environmental issues.

From 1979 until his retirement in 2004, Groth was a scientific expert at Consumers Union, publisher of Consumer Reports magazine. He is the author or coauthor of many books and studies, and he has also served on the Food Forum of the National Academy of Sciences and on expert committees for the World Health Organization and the UN Food and Agriculture Organization. This lecture is sponsored by UCSC Chapter of Sigma Xi (the Scientific Research Society), and by the UCSC Departments of Physics and of Microbiology and Environmental Toxicology (ETOX) . MethylmercuryMethylmercury PoisoningPoisoning inin High-EndHigh-End FishFish Consumers:Consumers: AA RiskRisk CommunicationCommunication ChallengeChallenge

Edward Groth III, PhD Groth Consulting Services Pelham, NY 10803 USA April, 2009 TopicsTopics toto bebe covered:covered:

 Background & context  Mercury in US fish and seafood  Summary of 24 case histories  Is it really methylmercury poisoning?  Dose-response issues  Fish involved in these cases  How prevalent a problem?  Research needs  Risk communication aspects ContextContext

 Americans are eating more fish, which benefits public health significantly, overall  But it also increases the likelihood of exposure to methylmercury, from eating fish  Risk is greater for people who eat a lot of fish  The type(s) of fish consumed also matter  Methylmercury exposure in general and extreme high-end exposure are each likely to increase if more Americans eat more fish ConventionalConventional HgHg Wisdom:Wisdom:

 CriticalCritical effecteffect == developmentaldevelopmental neurotoxicityneurotoxicity  PopulationsPopulations atat riskrisk == fetusesfetuses (i.e.(i.e. womenwomen ofof childbearingchildbearing age)age) andand youngyoung childrenchildren  NoNo appreciableappreciable riskrisk toto otherother populationspopulations  BenefitsBenefits (lower(lower risksrisks ofof CHDCHD && stroke)stroke) farfar outweighoutweigh HgHg risksrisks forfor generalgeneral populationpopulation ThisThis perspectiveperspective isis reflectedreflected asas recentlyrecently asas inin thethe 20062006 NAS/IOMNAS/IOM reportreport onon benefitsbenefits andand risksrisks ofof fishfish && seafoodseafood consumptionconsumption BasisBasis forfor C.W.:C.W.:

 EpidemiologyEpidemiology fromfrom incidentsincidents inin JapanJapan andand Iraq,Iraq, mostmost studiesstudies 30-4030-40 yearsyears agoago  FoundFound clear-cutclear-cut neurotoxicneurotoxic effectseffects inin adultsadults onlyonly atat highhigh dosesdoses (blood(blood HgHg >> ~200~200 ppb)ppb)  SomeSome effectseffects inin childrenchildren @@ >> 5050 ppbppb  AmountsAmounts ofof MeHgMeHg fromfrom fishfish inin ““normalnormal dietdiet”” believedbelieved toto bebe belowbelow thethe levellevel ofof concern,concern, exceptexcept forfor potentialpotential forfor fetalfetal exposureexposure KeyKey questions:questions:

 DoDo wewe needneed toto revisitrevisit andand updateupdate thisthis riskrisk assessment?assessment?  IfIf so,so, howhow mightmight wewe approachapproach thatthat task?task?  WhatWhat aboutabout ““abnormalabnormal”” (high-fish)(high-fish) diets?diets?  OnOn whatwhat issuesissues dodo wewe needneed betterbetter data?data?  GivenGiven whatwhat wewe knowknow andand dondon’’tt know,know, whatwhat adviceadvice shouldshould wewe givegive consumers?consumers? AA FewFew BasicBasic PrinciplesPrinciples ofof EnvironmentalEnvironmental HealthHealth RiskRisk isis aa ContinuumContinuum SensitivitySensitivity toto toxictoxic effectseffects variesvaries alongalong aa distributiondistribution

Number of individuals affected

Dose SensitiveSensitive subpopulationsubpopulation

Number affected

Dose FishFish consumptionconsumption andand methylmercurymethylmercury exposureexposure FishFish consumptionconsumption

 Long-termLong-term trendtrend ofof increasingincreasing perper capitacapita consumptionconsumption inin USUS  RecentRecent yearsyears at/nearat/near all-timeall-time highhigh  PatternsPatterns ofof consumptionconsumption alsoalso changingchanging  MoreMore freshfresh andand frozenfrozen steakssteaks andand filletsfillets  MoreMore (mostly(mostly imported)imported) shrimpshrimp  LessLess cannedcanned andand breaded/processedbreaded/processed fishfish USUS PerPer capitacapita fishfish consumption,consumption, pounds/year,pounds/year, 1990-20061990-2006 (NMFS)(NMFS)

90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 ConsumptionConsumption ofof selectedselected items,items, pounds/person/year,pounds/person/year, 1990-20061990-2006

90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 PerPer capitacapita consumption,consumption, cannedcanned fish,fish, pounds/year,pounds/year, 1990-20061990-2006

90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 TopTop 1010 SeafoodsSeafoods,, 2005-20072005-2007 US consumption in pounds per capita per year (NFI)

Rank 2005 2006 2007 . Species Lbs Species Lbs Species Lbs 1 Shrimp 4.10 Shrimp 4.40 Shrimp 4.10 2 Tuna, can 3.10 Tuna, can 2.90 Tuna, can 2.70 3 Salmon 2.43 Salmon 2.03 Salmon 2.36 4 Pollock 1.47 Pollock 1.64 Pollock 1.73 5 Catfish 1.03 Tilapia 1.00 Tilapia 1.14 6 Tilapia 0.85 Catfish 0.97 Catfish 0.88 7 Crab 0.64 Crab 0.66 Crab 0.68 8 Cod 0.57 Cod 0.51 Cod 0.47 9 Clams 0.44 Clams 0.44 Clams 0.45 10 Flatfish 0.37 Scallops 0.31 Flatfish 0.32 Total, Top 10 15.0 14.9 14.8 WhereWhere’’ss thethe mercury?mercury?

 AmongAmong popularpopular fishfish andand seafoodseafood choices,choices, howhow muchmuch doesdoes eacheach varietyvariety contributecontribute toto potentialpotential methylmercurymethylmercury exposure?exposure?  WhichWhich fishfish areare likelylikely toto contributecontribute mostmost toto methylmercurymethylmercury intake,intake, amongamong peoplepeople whowho eateat aa greatgreat dealdeal ofof fish?fish? MethylmercuryMethylmercury Exposure:Exposure: SourceSource StrengthsStrengths

 ContributionsContributions ofof differentdifferent fishfish andand seafoodseafood itemsitems toto totaltotal amountamount ofof mercurymercury inin thethe USUS fish/seafoodfish/seafood supply,supply, calculatedcalculated using:using: A:A: 20062006 USUS marketmarket datadata fromfrom NMFSNMFS B:B: MercuryMercury contentcontent fromfrom FDAFDA databasedatabase HgHg InputInput == (%(% ofof market)market) xx (Hg(Hg ppm)ppm) RelativeRelative HgHg ContributionsContributions

 Hg inputs calculated for 51 types of fish and shellfish for which there are both NMFS market data and FDA Hg data  Results are not precise indicators of exposure, but provide relative comparisons  Results can be ranked and compared various ways (e.g., percent of total Hg) AA KeyKey Fact:Fact:

The weighted average methylmercury concentration in the US seafood supply is 0.086 ppm Color-codingColor-coding fishfish forfor methylmercurymethylmercury contentcontent

 GREEN = very low = < 0.043 ppm  BLUE = below average = 0.044 - 0.086 ppm  BLACK = above average = 0.087 - 0.172 ppm  ORANGE = moderately high = 0.173 - 0.344 ppm  RED = high = 0.345 - 0.688 ppm  VIOLET = very high = > 0.688 ppm Note: Different breakpoints than FDA has used TopTop 1010 SeafoodsSeafoods,, 2005-20072005-2007 US consumption in pounds per capita per year (NFI)

Market Fish Share (%) ppm Hg Percent Hg

Tuna, all types 16.44 next slide 37.37 Haddock & Hake 4.86 0.170 9.73 Swordfish 0.44 0.976 5.06 Catfish 5.71 0.068 4.66 Cod 3.36 0.115 4.55 American lobster 1.22 0.310 4.46 Pollock 7.32 0.049 4.23 Shrimp 22.21 0.012 3.14 Salmon 6.83 0.028 2.25 Sea Bass 0.51 0.301 1.81

Total 77.26 Tuna,Tuna, byby typetype

Type Market % ppm Hg % Hg

Canned albacore 3.81 0.353 15.85

Canned light 11.41 0.118 15.86

Fresh/Frozen 1.22 0.384 5.66

Totals 16.44 37.37

(Insufficient supply data to specify contributions by tuna type to fresh/frozen category, e.g., bluefin, albacore, bigeye, etc.) CommentsComments onon TopTop 1010

 Swordfish is the only Violet (very high Hg) fish among the Top 10 sources  Two Green (very low Hg) and two Blue (below average Hg) items unlikely to be hazards; in Top 10 due to huge volume consumed  Two Black and two Orange items could lead to excessive exposure if eaten frequently  Tuna (two Red, one Black) is overwhelmingly the largest source  Top 10 account for more than _ of all mercury OtherOther ItemsItems ofof InterestInterest

Fish Market % ppm Hg % Hg Rank

Gulf Tilefish 0.01 1.450 0.171 40 Shark 0.07 0.988 0.815 21 King mackerel 0.05 0.730 0.430 29 Orange roughy 0.20 0.550 1.296 16 Marlin 0.02 0.489 0.115 42 Grouper 0.27 0.460 1.463 13 Bluefish 0.06 0.337 0.240 35 Snapper 0.86 0.137 1.388 15 Anchovies 3.06 0.050 1.803 11 Squid 1.92 0.070 1.583 12 Clams 2.04 0.023 0.553 28 Scallops 1.46 0.023 0.396 30 InterpretingInterpreting thesethese data:data:

 Tuna contributes 6 times as much mercury to potential US exposure as do swordfish, shark, Gulf tilefish and king mackerel combined.  Americans eat 29 times as much tuna as they eat of the four highest-mercury fish combined  Lobster, sea bass, cod, haddock and hake are more important sources than many varieties with higher mercury levels, due to market share  Two-thirds of the market is in the Green and Blue categories,categories, i.e.,i.e., lowlow mercurymercury MercuryMercury IntensityIntensity ofof CategoriesCategories

Weighted Intensity Category Mean Hg % Market % Hg Index .

Very Low 0.018 42.86 9.074 0.21

Below Avg 0.056 24.13 15.984 0.66

Above Avg 0.129 22.51 34.303 1.52

Mod. High 0.289 2.81 9.565 3.43

High 0.375 5.57 24.599 4.57

Very High 0.964 0.57 6.475 10.83 MercuryMercury IntensityIntensity IndicesIndices

 Are ratios, % mercury / % of market  Indicate the relative mercury dose a consumer ingests by eating an item from each category  Span a range of over 50-fold; i.e., fish in the VioletViolet category deliver over 50 times as much mercury, on average, as fish or shellfish in the GreenGreen category ““TroubleTrouble”” Scenarios:Scenarios:

WaysWays toto getget excessiveexcessive mercurymercury doses:doses: A. Eat very high Hg fish more often than rarely B. Eat moderately high or high Hg fish fairly often, i.e. once a week or more C. Eating above average Hg fish very often, i.e. twice a week or more, with occasional meals from categories in A or B There are large numbers of Americans (though a small percentage) with each of these consumption patterns CaseCase HistoriesHistories ofof methylmercurymethylmercury poisoningpoisoning inin peoplepeople whowho eateat aa lotlot ofof fishfish

OverOver TheThe LimitLimit  II wrotewrote itit forfor thethe MercuryMercury PolicyPolicy ProjectProject  PrimaryPrimary goal:goal: ToTo putput aa humanhuman faceface onon abstractabstract riskrisk conceptsconcepts  Sources:Sources: PublishedPublished casecase reports,reports, aa fewfew inin scientificscientific journals,journals, mostmost inin otherother mediamedia  II readilyreadily foundfound 2424 casescases ofof high-endhigh-end fishfish eaterseaters withwith methylmercurymethylmercury poisoningpoisoning  OnceOnce II hadhad thesethese data,data, II subjectedsubjected themthem toto somesome scientificscientific analysisanalysis CriteriaCriteria forfor InclusionInclusion

 SymptomsSymptoms consistentconsistent withwith methylmercurymethylmercury poisoningpoisoning  PatientPatient oftenoften consumedconsumed high-Hghigh-Hg fishfish  MeHgMeHg toxicitytoxicity diagnoseddiagnosed byby aa physicianphysician  SomeSome supportingsupporting datadata (e.g.,(e.g., bloodblood Hg)Hg)  PatientPatient stoppedstopped eatingeating high-Hghigh-Hg fishfish andand symptomssymptoms resolvedresolved  MostMost ofof thethe casescases meetmeet allall thesethese criteriacriteria WeaknessesWeaknesses inin thethe datadata

 MostMost casescases notnot peer-reviewedpeer-reviewed (only(only 44 ofof 2424 publishedpublished inin scientificscientific journals)journals)  SymptomsSymptoms areare generallygenerally subjectivesubjective  WideWide rangerange inin severityseverity ofof symptomssymptoms  ExposureExposure datadata (blood,(blood, hairhair Hg)Hg) unavailableunavailable inin somesome cases,cases, qualitativequalitative inin somesome othersothers  FishFish intakeintake basedbased onon patientpatient recallrecall  SomeSome patientspatients lostlost toto follow-upfollow-up FarFar fromfrom ideal:ideal:

 IndividualIndividual casecase historieshistories areare thethe ““lowestlowest”” formform ofof epidemiologicalepidemiological evidenceevidence  SomeSome ofof thesethese casescases areare fairlyfairly anecdotal,anecdotal, limitinglimiting confidenceconfidence inin theirtheir reliabilityreliability  But:But: LimitedLimited datadata areare nonethelessnonetheless data.data. WhatWhat cancan wewe learnlearn fromfrom thesethese cases?cases?  SomeSome provocativeprovocative observationsobservations emergeemerge fromfrom studystudy ofof thisthis limitedlimited datadata setset CriticalCritical Questions:Questions:

 WhoWho isis atat risk?risk?  IsIs itit reallyreally methylmercurymethylmercury poisoning?poisoning?  WhatWhat dosesdoses areare associatedassociated withwith harm?harm?  WhatWhat fishfish diddid thethe casescases eat?eat?  HowHow manymany otherother casescases mightmight therethere be?be?  WhatWhat researchresearch isis needed?needed?  WhatWhat adviceadvice shouldshould suchsuch high-endhigh-end fishfish consumersconsumers bebe getting?getting? WhoWho isis atat risk?risk?

 CasesCases werewere generallygenerally middle-agedmiddle-aged adults,adults, agesages 4040 toto 6666 atat diagnosisdiagnosis  FourFour casescases werewere childrenchildren  2020 ofof 2424 casescases ateate commercially-caughtcommercially-caught fish;fish; 44 werewere sportsport anglersanglers  TheThe 1616 adultsadults inin thethe formerformer groupgroup werewere allall health-conscioushealth-conscious individuals,individuals, tryingtrying toto eateat aa healthyhealthy diet,diet, equallyequally divideddivided byby gendergender

AA minorityminority ofof aa minorityminority

 NotNot ““typicaltypical”” Americans;Americans; realreal ““fishfish loverslovers””  MostMost probablyprobably areare aboveabove thethe 9999th,, somesome aboveabove 99.999.9th percentilepercentile ofof fishfish consumersconsumers  WithinWithin thatthat ““extremeextreme”” group,group, theythey preferprefer toto eateat higher-mercury,higher-mercury, predatorypredatory fish:fish: TunaTuna,, swordfishswordfish,, halibuthalibut,, seasea bassbass,, othersothers  SomeSome maymay alsoalso bebe moremore sensitivesensitive thanthan averageaverage toto toxictoxic effectseffects HowHow dodo wewe knowknow itit’’ss methylmercurymethylmercury poisoning?poisoning? SymptomsSymptoms seenseen inin cases:cases: Symptoms,Symptoms, continuedcontinued MethylmercuryMethylmercury poisoning?poisoning?

 SymptomsSymptoms matchmatch classicclassic symptomssymptoms  DiagnosedDiagnosed byby aa physicianphysician basedbased onon symptomssymptoms andand elevatedelevated blood/hairblood/hair HgHg  WhenWhen stoppedstopped eatingeating high-Hghigh-Hg fish,fish, bloodblood HgHg droppeddropped && symptomssymptoms resolvedresolved  NoNo evidenceevidence forfor otherother causescauses detecteddetected inin often-extensiveoften-extensive diagnosticdiagnostic processprocess  BottomBottom line:line: ItIt isis whatwhat itit appearsappears toto bebe HowHow suresure areare we?we?

 AbsoluteAbsolute proofproof isis nevernever possiblepossible  SeeSee detailsdetails inin thethe 2424 individualindividual casescases describeddescribed inin OverOver TheThe LimitLimit  SomeSome casescases areare aa bitbit questionablequestionable  ButBut thethe majoritymajority areare quitequite unequivocal:unequivocal: therethere isis virtuallyvirtually nono doubtdoubt thatthat thethe personperson gotgot mercurymercury poisoningpoisoning fromfrom eatingeating largelarge amountsamounts ofof fishfish withwith elevatedelevated HgHg contentcontent Dose-responseDose-response issues:issues:

 No quantitative blood level available in 3 cases with the most severe symptoms  Six cases with the mildest symptoms, no blood Hg available in 4, average 8 ppb in other two  But: No symptoms in one patient with highest quantified blood Hg (228 ppb)  Moderate to severe symptoms in 6 cases with blood Hg levels of 58-125 ppb  And: Similar moderate to severe symptoms in 8 other cases with blood Hg of 12-38 ppb Dose-responseDose-response && gender:gender:

2424 cases:cases: 2020 adults,adults, 44 childrenchildren

 MildMild symptoms:symptoms: 66 cases,cases, 55 malesmales andand oneone child,child, gendergender notnot specifiedspecified  ModerateModerate symptoms:symptoms: 1414 casescases • 5 males (3 adults, 2 children), avg bHg 68.4 ppb • 9 females (8 adults, 1 child), avg bHg 44.25 ppb  SevereSevere symptoms:symptoms: 33 cases,cases, allall malesmales Interpretations:Interpretations:

 AA smallsmall datadata set,set, butbut widewide differencesdifferences inin individualindividual sensitivitysensitivity toto toxictoxic effectseffects areare stillstill evidentevident  SensitiveSensitive individualsindividuals (1/3(1/3 ofof cases)cases) showshow symptomssymptoms atat bloodblood HgHg levelslevels longlong judgedjudged withoutwithout appreciableappreciable riskrisk (i.e.,(i.e., 12-3812-38 ppb)ppb)  MenMen seemseem moremore likelylikely toto experienceexperience eithereither severesevere oror mildmild symptomssymptoms  WomenWomen experiencedexperienced moderate-to-severemoderate-to-severe symptomssymptoms atat lowerlower dosesdoses thanthan menmen Low-doseLow-dose effects?effects?

 FrankFrank neurotoxicneurotoxic effectseffects associatedassociated inin somesome casescases herehere withwith farfar lowerlower exposureexposure levelslevels thanthan previouslypreviously recognizedrecognized  PossiblyPossibly hyper-sensitivehyper-sensitive individualsindividuals  ClinicalClinical toxicitytoxicity maymay bebe veryvery rarerare atat thesethese doses,doses, oror perhapsperhaps justjust rarelyrarely diagnoseddiagnosed  ButBut adverseadverse effectseffects atat lowlow dosesdoses areare notnot entirelyentirely unprecedentedunprecedented oror unexpectedunexpected Low-doseLow-dose effectseffects

 CartaCarta etet al.,al., 20052005 (Italy):(Italy):  2222 menmen whowho frequentlyfrequently ateate tuna,tuna, hadhad anan averageaverage bloodblood HgHg levellevel ofof 41.541.5 ppbppb  2222 controls,controls, hadhad averageaverage bHgbHg ofof 2.62.6 ppbppb  NeurobehavioralNeurobehavioral teststests ofof vigilance,vigilance, handhand tremor,tremor, psychomotorpsychomotor functionfunction  CasesCases performedperformed significantlysignificantly worseworse onon threethree functionalfunctional teststests (&(& worseworse onon allall 10)10) Low-doseLow-dose effects,effects, contcont’’dd

 YokooYokoo etet al.,al., 20032003 (Brazil):(Brazil):  BatteryBattery ofof neurobehavioralneurobehavioral andand cognitivecognitive teststests givengiven toto 129129 AmazonianAmazonian villagersvillagers  Adults,Adults, classifiedclassified byby exposureexposure basedbased onon hairhair HgHg levellevel (mean(mean 4.24.2 ++ 2.42.4 ppm,ppm, rangerange 0.560.56 -13.6-13.6 ppm)ppm)  Dose-relatedDose-related effectseffects ofof HgHg onon finefine motormotor speed,speed, dexterity,dexterity, concentrationconcentration andand somesome aspectsaspects ofof verbalverbal learninglearning && memorymemory ExposureExposure inin thisthis group:group:

 MeanMean hairhair mercurymercury ofof 4.24.2 ppmppm vs.vs. meanmean ofof aboutabout 11 ppmppm forfor USUS adultsadults  FourFour casescases inin OverOver TheThe LimitLimit hadhad hairhair HgHg levelslevels ofof 9,9, 12,12, 1313 andand 6868 ppmppm  I.e.,I.e., testedtested AmazonAmazon villagersvillagers havehave mercurymercury exposuresexposures notnot unlikeunlike AmericansAmericans whowho eateat aa lotlot ofof relativelyrelatively high-Hghigh-Hg fishfish Low-doseLow-dose effects,effects, contcont’’dd

 OkenOken etet al.,al., 2005,2005, 20082008 (Boston):(Boston):  CognitiveCognitive andand neurobehavioralneurobehavioral teststests inin infantsinfants && 3-yr-olds3-yr-olds vs.vs. maternalmaternal fishfish intakeintake  HighHigh fishfish consumptionconsumption correlatedcorrelated withwith improvedimproved cognitivecognitive performanceperformance  But:But: HighHigh mercurymercury exposureexposure correlatedcorrelated withwith decreaseddecreased cognitivecognitive performanceperformance  I.e.,I.e., antagonisticantagonistic effectseffects OkenOken etet al.al.’’ss subjects:subjects:

 ““HighHigh fish-eatersfish-eaters”” consumedconsumed onlyonly twotwo fishfish mealsmeals perper weekweek (>(> twicetwice USUS average)average)  HighHigh mercurymercury exposureexposure == >> 9090th percentilepercentile w/inw/in group,group, == hairhair HgHg >> 1.21.2 ppmppm  9090th percentilepercentile forfor bloodblood HgHg inin womenwomen inin NortheastNortheast USUS (NHANES)(NHANES) == 5.25.2 ppbppb  Inference:Inference: AdverseAdverse HgHg effectseffects onon thethe fetalfetal brainbrain maymay occuroccur @@ >> 55 ppbppb maternalmaternal bHgbHg ConfirmingConfirming StudiesStudies

 Lederman et al. (2008), New York City; mean maternal blood Hg level 2.29 ppb  Jedrychowski et al. (2006), Krakow, Poland; mean maternal blood Hg 0.75 ppb  Davidson et al. (2008), Seychelles; mean maternal hair Hg 5.7 ppm (NOTE: Previous reports from Seychelles had failed to see effects; confounding by nutritional benefits of fish consumption) Conclusions:Conclusions:

 WeWe areare approachingapproaching aa pointpoint wherewhere ourour viewview ofof low-doselow-dose methylmercurymethylmercury effectseffects maymay undergoundergo radicalradical revision,revision, asas occurredoccurred forfor leadlead toxicitytoxicity aroundaround 1979-80.1979-80.  Sub-clinicalSub-clinical effectseffects measuredmeasured byby sensitivesensitive teststests areare likelylikely toto bebe farfar moremore widespreadwidespread thanthan overtovert illnessillness BACKBACK TOTO OUROUR 2424 CASESCASES…… WhatWhat fishfish diddid theythey eat?eat? (commercially-caught(commercially-caught fish,fish, 2121 cases)cases)

20 18 16 14 12 10 8 6 4 2 0 Tuna, all Swordfish Halibut Sea BassYellowtail King types Mackerel OnlyOnly sixsix FishFish VarietiesVarieties involvedinvolved inin thesethese 2121 casescases

 TunaTuna (all(all types):types): 1818 cases,cases, 8686 %%  SwordfishSwordfish:: 88 cases,cases, 3838 %%  HalibutHalibut:: 33 cases,cases, 1414 %%  SeaSea bassbass:: 33 cases,cases, 1414 %%  YellowtailYellowtail:: 22 cases,cases, 10%10%  KingKing mackerelmackerel:: 11 case,case, 55 %% (> 100% because many cases ate more than one type of high-mercury fish) Noteworthy:Noteworthy:

 TwoTwo ofof thethe ““troubletrouble scenariosscenarios”” applyapply herehere  SomeSome patientspatients ateate swordfishswordfish,, aa veryvery highhigh HgHg fish,fish, oftenoften  ButBut thethe majoritymajority ateate TunaTuna,, seasea bassbass,, halibuthalibut andand yellowtailyellowtail,, allall fishfish withwith lessless extremeextreme HgHg levelslevels  TunaTuna waswas aa sourcesource inin aa largelarge majoritymajority ofof thethe casescases andand waswas thethe onlyonly knownknown sourcesource inin 99 casescases (43%)(43%) MercuryMercury LevelsLevels inin CommerciallyCommercially CaughtCaught FishFish InvolvedInvolved inin CasesCases Fish # of Cases ppm Hg Tuna, fresh/frozen 11 0.384 Swordfish 8 0.976 Tuna, canned, type not specified 4 0.118 Tuna, canned, albacore 3 0.353 Tuna, sushi 3 0.10-2.76 Halibut 3 0.220 Sea bass 2 0.301 Yellowtail 2 0.484 Tuna, bluefin 1 ~1.0 Sea bass, Chilean 1 0.600 King mackerel 1 0.730 . Most data from US FDA; tuna sushi, NY Times & Houston Chronicle; bluefin estimated from sushi data; Chilean sea bass, Knobeloch et al. (2005); Yellowtail, FL Fish & Wildlife Commission (2003) Summary:Summary:

 One-thirdOne-third ofof casescases (8(8 patients)patients) ateate aa high-high- mercurymercury fishfish ((swordfishswordfish)) repeatedlyrepeatedly  OneOne childchild casecase ateate somesome kingking mackerelmackerel,, butbut alsoalso ateate aa lotlot ofof cannedcanned tunatuna  TheThe largelarge majoritymajority ofof casescases ateate mostlymostly moderatelymoderately highhigh andand highhigh mercurymercury fish:fish: tunatuna (fresh/frozen(fresh/frozen steaks,steaks, canned,canned, andand sushi),sushi), halibuthalibut,, seasea bassbass andand yellowtailyellowtail  NineNine casescases (43%)(43%) ateate onlyonly tunatuna IfIf therethere werewere aa signsign aboveabove mymy desk,desk, herehere’’ss whatwhat itit mightmight say:*say:*

* with apologies to James Carville ItIt’’ss thethe tuna,tuna, stupid!stupid! HowHow manymany casescases mightmight bebe ““outout therethere””??

PossiblePossible sizesize ofof populationpopulation atat riskrisk estimatedestimated byby threethree differentdifferent methods:methods:

 Back-of-the-envelopeBack-of-the-envelope  InferencesInferences fromfrom publishedpublished studiesstudies  InferencesInferences fromfrom NHANESNHANES datadata ““ExtremeExtreme”” FishFish EatersEaters

FDA estimates:  Population Average Fish Consumption: Women: 14.3 g/day Men: 18.6 g/day  99th Percentile of Fish Consumption: Women: 95 g/day Men: 134 g/day If a typical serving is 150-180 grams (more for men), 99th percentile eats fish ~ 4 to 5 times per week Back-of-the-envelopeBack-of-the-envelope

 Assume:Assume: ExtremeExtreme fish-eatersfish-eaters areare aboveabove thethe 9999th percentilepercentile inin fishfish consumptionconsumption  Assume:Assume: 0.10.1 toto 1010 percentpercent repeatedlyrepeatedly eateat high-mercuryhigh-mercury fishfish 3,250,0003,250,000 consumersconsumers xx (0.1(0.1 toto 1010 percent)percent) == 3,2503,250 toto 325,000325,000 possiblepossible casescases LimitationsLimitations ofof BOTEBOTE method:method:

 Cases might occur below 99th percentile; i.e., ours varied from <1 to >10 fish meals per week  Very few data from which to estimate reliably how many people repeatedly eat higher-Hg fish; wide range of uncertainty (and perhaps >10% repeatedly eat tuna?)  Serving size, specific type of fish also matter  Method estimates only exposure; can’t say what fraction of people with high-end exposure might experience symptoms PublishedPublished StudiesStudies CarringtonCarrington && BolgerBolger (2003)(2003)  MaximumMaximum assumedassumed fishfish intakeintake == 1818 ozoz perper weekweek (( == << 99th99th percentile)percentile)  EstimatedEstimated 9999th percentilepercentile baselinebaseline bHgbHg inin womenwomen ofof childbearingchildbearing ageage == 16.116.1 ppb,ppb, andand 99.999.9th percentilepercentile bHgbHg == 26.326.3 ppbppb  I.e.,I.e., 99.999.9th percentilepercentile consumerconsumer (1(1 inin 1,0001,000 people)people) hashas bloodblood HgHg inin thethe low-midlow-mid rangerange observedobserved inin casescases inin OverOver TheThe LimitLimit RepeatRepeat consumptionconsumption data:data:

 Carrington & Bolger also have estimated the frequency of repeat consumption from NHANES data  About 10 percent of women choose the same fish > 80% of the time  Problems: Too few data to estimate freq. of repeat eating of low-market share high mercury fish; & data are just for women InferencesInferences fromfrom C&BC&B model:model:

 Roughly 1 in 1,000 consumers may have blood Hg levels in the range associated with toxic symptoms in sensitive individuals among the 24 cases (i.e., > 20 ppb)  For a lower exposure level (e.g., 15 ppb), the number possibly at risk may rises to 2 in 1000  Far less) than 1 in 1,000 have bHg levels above, say, 50 ppb Limitations:Limitations:

 AppliesApplies toto womenwomen ofof childbearingchildbearing ageage  ModelModel lackslacks empiricalempirical datadata onon thosethose (rare)(rare) individualsindividuals whowho repeatedlyrepeatedly choosechoose higher-higher- mercurymercury fishfish  ReliedRelied onon NHANESNHANES fishfish consumptionconsumption data;data; NHANESNHANES samplesample isis nationallynationally balanced,balanced, doesdoes notnot includeinclude manymany membersmembers ofof ethnicethnic oror tribaltribal minoritiesminorities withwith high-fishhigh-fish dietsdiets PublishedPublished epiepi studiesstudies

 VeryVery fewfew publishedpublished epidemiologicalepidemiological datadata  HightowerHightower && MooreMoore (2003):(2003): 720720 patients,patients, ~100~100 withwith elevatedelevated bloodblood HgHg (>(> 55 ppb),ppb), ~~ 55 hadhad symptomssymptoms (case(case raterate == 0.7%)0.7%)  KnobelochKnobeloch etet al.al. (2005),(2005), 20002000 volunteers;volunteers; 77 casescases w.w. elevatedelevated bloodblood HgHg (0.35%);(0.35%); 33 withwith symptomssymptoms (0.15%)(0.15%)  Non-representativeNon-representative populationspopulations inin eacheach case.case. ProjectedProjected incidenceincidence thusthus << 0.1%0.1% InferencesInferences fromfrom NHANESNHANES

 MeasuredMeasured bloodblood HgHg inin 5,2145,214 womenwomen andand children,children, 1999-20041999-2004  NoNo adultadult men,men, nono olderolder women,women, notnot fullyfully balancedbalanced regionallyregionally oror ethnicallyethnically  MaximumMaximum bloodblood HgHg levellevel inin thethe NHANESNHANES samplesample waswas 3333 ppbppb  WhatWhat doesdoes thisthis telltell us?us? LevelsLevels aboveabove 3333 ppb?ppb?

 AnalysisAnalysis ofof statisticalstatistical powerpower ofof sample:sample:  ConsiderConsider aa highhigh bloodblood HgHg level,level, defineddefined herehere (arbitrarily)(arbitrarily) asas >> 3333 ppb.ppb.  TheThe NHANESNHANES samplesample includedincluded nono oneone withwith aa levellevel thatthat highhigh  HowHow manymany peoplepeople inin thethe USUS populationpopulation ofof 325325 millionmillion couldcould havehave levelslevels higherhigher thanthan that,that, andand NHANESNHANES wouldwould stillstill bebe unlikelyunlikely toto includeinclude atat leastleast oneone ofof them?them? Probabilities:Probabilities:

 AssumeAssume forfor thisthis exerciseexercise thatthat thethe NHANESNHANES samplesample waswas randomrandom andand representativerepresentative ofof thethe USUS asas aa wholewhole  IfIf thethe incidenceincidence ofof bHgbHg >> 3333 ppbppb werewere 11 inin 1,0001,000 people,people, thethe probabilityprobability thatthat NHANESNHANES wouldwould includeinclude zerozero isis (0.999)(0.999)5214 == 0.00540.0054  IfIf thethe incidenceincidence ofof bHgbHg >> 3333 ppbppb werewere 11 inin 10,00010,000 people,people, thethe probabilityprobability thatthat NHANESNHANES wouldwould includeinclude zerozero isis (0.9999)(0.9999)5214 == 0.590.59 WithWith 95%95% Confidence:Confidence:

 IfIf thethe incidenceincidence werewere 11 inin 1,7421,742 people,people, thethe probabilityprobability thatthat NHANESNHANES wouldwould notnot includeinclude anyany isis 0.05.0.05.  I.e.,I.e., wewe cancan bebe 95%95% confidentconfident thatthat therethere areare nono moremore thanthan 186,567186,567 (325,000,000(325,000,000 ÷÷ 1,742)1,742) peoplepeople inin thethe USUS withwith bloodblood HgHg >> 3333 ppb.ppb.  Or,Or, 0.060.06 percentpercent ofof thethe populationpopulation oror lessless areare likelylikely toto havehave levelslevels aboveabove 3333 ppb.ppb. Comments:Comments:

 This conclusion is not very reassuring  This analysis dealt with blood Hg levels above 33 ppb (max observed in NHANES sample)  The same probabilities apply to 34 ppb, 84 ppb and 134 ppb, say, but we know the incidence decreases sharply as blood Hg level increases  Cases suggest that symptoms may occur at 33 ppb or less in some sensitive patients  Sub-clinical effects on cognitive processes and fine-motor coordination are also a concern TriangulationTriangulation

 C&BC&B modelmodel suggestssuggests 0.10.1 percentpercent ofof USUS womenwomen couldcould havehave bHgbHg >> 2626 ppbppb  NHANESNHANES analysisanalysis suggestssuggests 0.060.06 percentpercent couldcould havehave bHgbHg >> 3333 ppbppb  NeitherNeither predictspredicts frequencyfrequency ofof symptomssymptoms  PublishedPublished studiesstudies suggestsuggest symptomssymptoms inin 0.150.15 –– 0.70.7 percentpercent ofof twotwo highlyhighly selectedselected populations;populations; generalgeneral incidenceincidence isis surelysurely less,less, butbut cancan’’tt saysay howhow muchmuch lessless BottomBottom lines:lines:  NoneNone ofof thesethese estimationestimation methodsmethods isis veryvery preciseprecise oror satisfactorysatisfactory  ButBut theythey convergeconverge aroundaround aa possiblepossible incidenceincidence ofof aboutabout 0.060.06 toto 0.10.1 percentpercent  I.e.,I.e., fromfrom 200,000200,000 toto 300,000300,000 AmericansAmericans maymay havehave elevatedelevated bloodblood HgHg (>(> ~25~25 ppb)ppb)  IncidenceIncidence ofof elevatedelevated bloodblood HgHg doesdoes notnot predictpredict thethe incidenceincidence ofof toxictoxic symptomssymptoms  ActualActual numbernumber ofof casescases couldcould thereforetherefore bebe (much)(much) smallersmaller (tens(tens ofof thousands?)thousands?) Comments:Comments:

 TheThe needneed toto narrownarrow thesethese uncertaintiesuncertainties byby focusedfocused researchresearch isis urgenturgent  Meanwhile,Meanwhile, however,however, wewe maymay wishwish toto actact asas ifif therethere couldcould bebe fromfrom severalseveral thousandthousand toto aa fewfew hundredhundred thousandthousand possiblepossible casescases ofof methylmercurymethylmercury poisoningpoisoning amongamong high-high- endend USUS fishfish consumersconsumers ResearchResearch needs:needs:

 MoreMore casecase historieshistories needneed toto bebe publishedpublished inin medicalmedical journalsjournals (I(I’’dd welcomewelcome referrals)referrals)  FocusedFocused studiesstudies usingusing sensitivesensitive outcomeoutcome measuresmeasures forfor methylmercurymethylmercury effectseffects onon thethe CNSCNS shouldshould bebe donedone onon peoplepeople whowho eateat aa greatgreat dealdeal ofof fishfish (adults(adults && kids)kids)  SimilarSimilar studiesstudies shouldshould bebe donedone onon aa largelarge cross-sectioncross-section ofof thethe population,population, stratifiedstratified byby HgHg exposureexposure MoreMore researchresearch needs:needs:

 BetterBetter datadata areare neededneeded onon high-Hghigh-Hg fishfish consumption:consumption: HowHow manymany peoplepeople eateat suchsuch fishfish repeatedly,repeatedly, andand howhow muchmuch dodo theythey eateat mod-highmod-high,, highhigh andand veryvery highhigh HgHg fish?fish?  BetterBetter datadata neededneeded onon HgHg levelslevels inin somesome fish,fish, includingincluding lowlow andand belowbelow averageaverage HgHg fish,fish, recommendedrecommended asas safersafer choiceschoices (FDA(FDA datadata quitequite sparsesparse inin manymany respects)respects) AdviceAdvice forfor ConsumersConsumers whowho eateat aa lotlot ofof fishfish

 Who: Population needing advice is not just mothers-to-be; anyone else who eats a lot of the wrong fish (> twice a week) may be at risk too  What fish: It’s not just very high Hg fish; high and moderately high fish also are clearly a problem if eaten often, and above average Hg fish can also contribute significantly to risk of excessexcess exposureexposure ifif eateneaten inin largelarge amountsamounts WhichWhich fishfish toto choose?choose?

 Fish and shellfish in the GreenGreen and BlueBlue categories are unlikely to lead to excess exposure no matter how much one eats  These two “safe” categories account for 67 percent of the market  So, motivated consumers can easily find low-mercury choices TopTop 1010 SeafoodsSeafoods,, 2005-20072005-2007 US consumption in pounds per capita per year (NFI)

““EatEat lotslots ofof low-low- mercurymercury fishfish”” HardHard toto getget thisthis right:right:

 ConflictingConflicting messagesmessages fromfrom variousvarious expertexpert sourcessources and/orand/or interestedinterested parties:parties:  NotNot right:right: ““BenefitsBenefits outweighoutweigh risks,risks, dondon’’tt worryworry aboutabout mercury.mercury.”” (False(False trade-off)trade-off)  NotNot right:right: ““EatEat lotslots ofof fish.fish.”” (Fails(Fails toto makemake importantimportant risk-relatedrisk-related distinctions.)distinctions.)  NotNot right:right: ““ToTo avoidavoid mercurymercury risk,risk, dondon’’tt eateat fish.fish.”” (Dismisses(Dismisses benefits.)benefits.) CommunicationCommunication challengeschallenges

 Americans consume a great deal of tuna  Some people also eat other moderately high, high, or very high Hg fish repeatedly  Need to advise those consumers as a distinct sub-population at significant risk  They need more and better advice about the mercury content of all popular fish and shellfish varieties, and improved guidance to choose low- mercury items One idea about what consumer advice might look like ConsumerConsumer AdviceAdvice

IfIf youyou eateat fishfish twicetwice aa weekweek oror lessless,, choosechoose fishfish asas follows:follows: GreenGreen oror BlueBlue:: AsAs oftenoften asas youyou likelike Black:Black: UpUp toto onceonce perper weekweek OrangeOrange oror RedRed:: UpUp toto once/2once/2 weeksweeks VioletViolet:: UpUp toto onceonce perper monthmonth ConsumerConsumer Advice,Advice, contcont’’dd

IfIf youyou eateat fishfish 3-43-4 timestimes aa week,week, choosechoose fishfish asas follows:follows: GreenGreen oror BlueBlue:: AsAs oftenoften asas youyou likelike Black:Black: UpUp toto onceonce inin twotwo weeksweeks OrangeOrange//RedRed:: UpUp toto onceonce perper monthmonth VioletViolet:: UpUp toto onceonce perper 33 monthsmonths MPPMPP Advice,Advice, contcont’’dd

IfIf youyou eateat fishfish 55 timestimes aa weekweek oror moremore,, choosechoose fishfish asas follows:follows: GreenGreen:: AsAs oftenoften asas youyou likelike BlueBlue:: UpUp toto onceonce aa weekweek Black:Black: UpUp toto onceonce aa monthmonth OrangeOrange//RedRed:: UpUp toto onceonce inin threethree monthsmonths VioletViolet:: OnceOnce oror twicetwice aa yearyear ModesModes ofof AdviceAdvice

 Government advisories  NGO & private sector reports & web  Point of sale information  Media articles

Effort is needed to improve information through all these modes & media